Adding lost bandicoot usage example by shrit · Pull Request #245 · mlpack/examples

shrit · 2025-02-22T15:24:59Z

No description provided.

Signed-off-by: Omar Shrit <omar@avontech.fr>

github-actions · 2025-02-22T15:25:10Z

👈 Launch a binder notebook on branch shrit/examples/coot_2

Signed-off-by: Omar Shrit <omar@avontech.fr>

zoq

Great example, I used it to verify the Vulkan backend, I left a few suggestions to make this example build by default against the CUDA backend, but to make it work against Vulkan or OpenCL you just need to uncomment the corresponding line the the Makefile.

I also commented in the inference part after the training is done. Also also added the accuracy function to output the accuracy after the training finished. That said, all backends have very similar loss curves and final loss as well.

Note, I needed to build against the bandicoot unstable branch to make this work (we have a few patches open as MR to get this working across all backends).

zoq · 2026-08-28T19:30:32Z

+coot::Row<size_t> getLabels(coot::mat predOut)
+{
+  coot::Row<size_t> predLabels(predOut.n_cols);
+  for (coot::uword i = 0; i < predOut.n_cols; ++i)
+  {
+    // predLabels(i) = predOut.col(i).index_max();


Suggested change

coot::Row<size_t> getLabels(coot::mat predOut)

{

coot::Row<size_t> predLabels(predOut.n_cols);

for (coot::uword i = 0; i < predOut.n_cols; ++i)

{

// predLabels(i) = predOut.col(i).index_max();

coot::Row<size_t> getLabels(const coot::mat& predOut)

{

coot::Row<size_t> predLabels(predOut.n_cols);

for (coot::uword i = 0; i < predOut.n_cols; ++i)

{

predLabels(i) = predOut.col(i).index_max();

zoq · 2026-08-28T19:31:24Z

+  }
+  return predLabels;
+}
+


Suggested change

double accuracy(const coot::Row<size_t>& predLabels, const coot::mat& labels)

{

const coot::Row<size_t> trueLabels = coot::conv_to<coot::Row<size_t>>::from(labels);

return coot::accu(predLabels == trueLabels) / (double) labels.n_elem * 100;

}

to measure accuracy

zoq · 2026-08-28T19:33:03Z

+  const coot::mat trainX =
+      coot::conv_to<coot::mat>::from(train.submat(1, 0, train.n_rows - 1, train.n_cols - 1) / 255.0);
+  const coot::mat validX =
+      coot::conv_to<coot::mat>::from(valid.submat(1, 0, valid.n_rows - 1, valid.n_cols - 1) / 255.0);


Suggested change

const coot::mat trainX =

coot::conv_to<coot::mat>::from(train.submat(1, 0, train.n_rows - 1, train.n_cols - 1) / 255.0);

const coot::mat validX =

coot::conv_to<coot::mat>::from(valid.submat(1, 0, valid.n_rows - 1, valid.n_cols - 1) / 255.0);

const coot::mat trainX = coot::conv_to<coot::mat>::from(

train.submat(1, 0, train.n_rows - 1, train.n_cols - 1) / 255.0);

const coot::mat validX = coot::conv_to<coot::mat>::from(

valid.submat(1, 0, valid.n_rows - 1, valid.n_cols - 1) / 255.0);

zoq · 2026-08-28T19:34:30Z

+  //coot::mat predOut;
+  //// Getting predictions on training data points.
+  //model.Predict(trainX, predOut);
+  //// Calculating accuracy on training data points.
+  //coot::Row<size_t> predLabels = getLabels(predOut);
+  //double trainAccuracy =
+      //coot::accu(predLabels == trainY) / (double) trainY.n_elem * 100;
+  //// Getting predictions on validating data points.
+  //model.Predict(validX, predOut);
+  //// Calculating accuracy on validating data points.
+  //predLabels = getLabels(predOut);
+  //double validAccuracy =
+      //coot::accu(predLabels == validY) / (double) validY.n_elem * 100;


Suggested change

//coot::mat predOut;

//// Getting predictions on training data points.

//model.Predict(trainX, predOut);

//// Calculating accuracy on training data points.

//coot::Row<size_t> predLabels = getLabels(predOut);

//double trainAccuracy =

//coot::accu(predLabels == trainY) / (double) trainY.n_elem * 100;

//// Getting predictions on validating data points.

//model.Predict(validX, predOut);

//// Calculating accuracy on validating data points.

//predLabels = getLabels(predOut);

//double validAccuracy =

//coot::accu(predLabels == validY) / (double) validY.n_elem * 100;

coot::mat predOut;

// Getting predictions on training data points.

model.Predict(trainX, predOut);

// Calculating accuracy on training data points.

double trainAccuracy = accuracy(getLabels(predOut), trainY);

// Getting predictions on validating data points.

model.Predict(validX, predOut);

// Calculating accuracy on validating data points.

double validAccuracy = accuracy(getLabels(predOut), validY);

zoq · 2026-08-28T19:34:47Z

+  //cout << "Accuracy: train = " << trainAccuracy << "%,"
+       //<< "\t valid = " << validAccuracy << "%" << endl;


Suggested change

//cout << "Accuracy: train = " << trainAccuracy << "%,"

//<< "\t valid = " << validAccuracy << "%" << endl;

cout << "Accuracy: train = " << trainAccuracy << "%,"

<< "\t valid = " << validAccuracy << "%" << endl;

zoq · 2026-08-28T19:36:02Z

+  //data::Save("model.bin", "model", model, false);
+
+  // Loading test dataset (the one whose predicted labels
+  // should be sent to kaggle website).
+  //data::Load("../data/mnist_test.csv", dataset, true);
+  //coot::mat testY = dataset.row(0);
+  //dataset.shed_row(0); // Strip labels before predicting.
+  //dataset /= 255.0; // Apply the same normalization as to the training data.
+
+  //cout << "Predicting on test set..." << endl;
+  //coot::mat testPredOut;
+  //// Getting predictions on test data points.
+  //model.Predict(dataset, testPredOut);
+  //// Generating labels for the test dataset.
+  //coot::Row<size_t> testPred = getLabels(testPredOut);
+
+  //double testAccuracy = coot::accu(testPred == testY) /
+      //(double) testY.n_elem * 100;
+  //cout << "Accuracy: test = " << testAccuracy << "%" << endl;
+
+  //cout << "Saving predicted labels to \"results.csv\" ..." << endl;
+  //testPred.save("results.csv", coot::csv_ascii);
+
+  //cout << "Neural network model is saved to \"model.bin\"" << endl;


Suggested change

//data::Save("model.bin", "model", model, false);

// Loading test dataset (the one whose predicted labels

// should be sent to kaggle website).

//data::Load("../data/mnist_test.csv", dataset, true);

//coot::mat testY = dataset.row(0);

//dataset.shed_row(0); // Strip labels before predicting.

//dataset /= 255.0; // Apply the same normalization as to the training data.

//cout << "Predicting on test set..." << endl;

//coot::mat testPredOut;

//// Getting predictions on test data points.

//model.Predict(dataset, testPredOut);

//// Generating labels for the test dataset.

//coot::Row<size_t> testPred = getLabels(testPredOut);

//double testAccuracy = coot::accu(testPred == testY) /

//(double) testY.n_elem * 100;

//cout << "Accuracy: test = " << testAccuracy << "%" << endl;

//cout << "Saving predicted labels to \"results.csv\" ..." << endl;

//testPred.save("results.csv", coot::csv_ascii);

//cout << "Neural network model is saved to \"model.bin\"" << endl;

// Loading test dataset (the one whose predicted labels

// should be sent to kaggle website).

data::Load("../../../data/mnist_test.csv", dataset, true);

const coot::mat testY = coot::conv_to<coot::mat>::from(dataset.row(0));

dataset.shed_row(0); // Strip labels before predicting.

dataset /= 255.0; // Apply the same normalization as to the training data.

const coot::mat testX = coot::conv_to<coot::mat>::from(dataset);

cout << "Predicting on test set..." << endl;

coot::mat testPredOut;

// Getting predictions on test data points.

model.Predict(testX, testPredOut);

double testAccuracy = accuracy(getLabels(testPredOut), testY);

cout << "Accuracy: test = " << testAccuracy << "%" << endl;

zoq · 2026-08-28T19:37:26Z

+# This is a simple Makefile used to build the example source code.
+# This example might requires some modifications in order to work correctly on
+# your system.
+# This example trains mlpack neural network on GPU via OpenCL or CUDA, and it uses


Suggested change

# This example trains mlpack neural network on GPU via OpenCL or CUDA, and it uses

# This example trains mlpack neural network on GPU via OpenCL, CUDA or Vulkan, and it uses

zoq · 2026-08-28T19:38:16Z

+LIBS_NAME := bandicoot openblas armadillo
+
+CXX := g++
+CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -g -DNDEBUG -fopenmp


Suggested change

CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -g -DNDEBUG -fopenmp

CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -DNDEBUG -fopenmp

# Select the bandicoot backend to run the example on.

CXXFLAGS += -DCOOT_DEFAULT_BACKEND=CUDA_BACKEND

# CXXFLAGS += -DCOOT_DEFAULT_BACKEND=CL_BACKEND

# CXXFLAGS += -DCOOT_DEFAULT_BACKEND=VULKAN_BACKEND

zoq · 2026-08-28T19:39:09Z

+CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -g -DNDEBUG -fopenmp
+# Use these CXXFLAGS instead if you want to compile with debugging symbols and
+# without optimizations.
+# CXXFLAGS += -std=c++14 -Wall -Wextra -g -O0


Suggested change

# CXXFLAGS += -std=c++14 -Wall -Wextra -g -O0

# CXXFLAGS += -std=c++17 -Wall -Wextra -g -O0

zoq · 2026-08-28T19:40:02Z

+INCLFLAGS := -I/meta/ensmallen/include
+# If you have mlpack or ensmallen installed somewhere nonstandard, uncomment and
+# update the lines below.
+INCLFLAGS += -I/opt/cuda/targets/x86_64-linux/include
+INCLFLAGS += -I/meta/mlpack/src
+#INCLFLAGS += -I/meta/m


Suggested change

INCLFLAGS := -I/meta/ensmallen/include

# If you have mlpack or ensmallen installed somewhere nonstandard, uncomment and

# update the lines below.

INCLFLAGS += -I/opt/cuda/targets/x86_64-linux/include

INCLFLAGS += -I/meta/mlpack/src

#INCLFLAGS += -I/meta/m

# If you have mlpack, ensmallen or bandicoot installed somewhere nonstandard,

# uncomment and update the lines below.

# INCLFLAGS += -I/path/to/mlpack/include/

# INCLFLAGS += -I/path/to/ensmallen/include/

# INCLFLAGS += -I/path/to/bandicoot/include/

# The CUDA backend of bandicoot needs the CUDA headers; comment this out if

# bandicoot was built without CUDA support.

INCLFLAGS += -I/usr/local/cuda/include

CXXFLAGS += $(INCLFLAGS)

Adding lost bandicoot usage example

a0d4851

Signed-off-by: Omar Shrit <omar@avontech.fr>

Fix some errors

c4ea7ac

Signed-off-by: Omar Shrit <omar@avontech.fr>

github-actions Bot added the s: stale label Mar 28, 2025

github-actions Bot closed this Apr 11, 2025

shrit added the s: keep open label May 5, 2025

shrit reopened this May 5, 2025

shrit added 2 commits September 24, 2025 21:07

Fix the layers and conv_to

fc26334

Signed-off-by: Omar Shrit <omar@avontech.fr>

Add more libs and fix the changes related to NN code base

6f3231e

Signed-off-by: Omar Shrit <omar@avontech.fr>

rcurtin mentioned this pull request Dec 15, 2025

Optimize NegativeLogLikelihood for Bandicoot mlpack/mlpack#4056

Merged

zoq reviewed Aug 28, 2026

View reviewed changes

Navigation Menu

Uh oh!

Adding lost bandicoot usage example - #245

Adding lost bandicoot usage example#245
shrit wants to merge 4 commits into
mlpack:masterfrom
shrit:coot_2

shrit commented Feb 22, 2025

Uh oh!

github-actions Bot commented Feb 22, 2025

Uh oh!

zoq left a comment

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

zoq Aug 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

+double accuracy(const coot::Row<size_t>& predLabels, const coot::mat& labels)
+{
+  const coot::Row<size_t> trueLabels = coot::conv_to<coot::Row<size_t>>::from(labels);
+  return coot::accu(predLabels == trueLabels) / (double) labels.n_elem * 100;
+}

		//cout << "Accuracy: train = " << trainAccuracy << "%,"
		//<< "\t valid = " << validAccuracy << "%" << endl;

	# This example trains mlpack neural network on GPU via OpenCL or CUDA, and it uses
	# This example trains mlpack neural network on GPU via OpenCL, CUDA or Vulkan, and it uses

-CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -g -DNDEBUG -fopenmp
+CXXFLAGS += -std=c++17 -Wall -Wextra -O3 -DNDEBUG -fopenmp
+# Select the bandicoot backend to run the example on.
+CXXFLAGS += -DCOOT_DEFAULT_BACKEND=CUDA_BACKEND
+# CXXFLAGS += -DCOOT_DEFAULT_BACKEND=CL_BACKEND
+# CXXFLAGS += -DCOOT_DEFAULT_BACKEND=VULKAN_BACKEND

	# CXXFLAGS += -std=c++14 -Wall -Wextra -g -O0
	# CXXFLAGS += -std=c++17 -Wall -Wextra -g -O0

-INCLFLAGS := -I/meta/ensmallen/include
-# If you have mlpack or ensmallen installed somewhere nonstandard, uncomment and
-# update the lines below.
-INCLFLAGS += -I/opt/cuda/targets/x86_64-linux/include
-INCLFLAGS += -I/meta/mlpack/src
-#INCLFLAGS += -I/meta/m
+# If you have mlpack, ensmallen or bandicoot installed somewhere nonstandard,
+# uncomment and update the lines below.
+# INCLFLAGS += -I/path/to/mlpack/include/
+# INCLFLAGS += -I/path/to/ensmallen/include/
+# INCLFLAGS += -I/path/to/bandicoot/include/
+# The CUDA backend of bandicoot needs the CUDA headers; comment this out if
+# bandicoot was built without CUDA support.
+INCLFLAGS += -I/usr/local/cuda/include
+CXXFLAGS += $(INCLFLAGS)

Uh oh!

Conversation

shrit commented Feb 22, 2025

Uh oh!

github-actions Bot commented Feb 22, 2025

Uh oh!

zoq left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants